Pymongo：如何确保仅从副本集中的所需主机（尤其是次要主机）进行连接和读取

Question

Background

mongoDB production cluster (not sharded) 3 hosts. They belong to replica set rs0

• H1 - Primary
• H2 - Secondary
• H3 - Secondary

I am trying to write a python script using pymongo . This being a production cluster, I want the script to read only from H3 secondary replica set so as to not burden primary and keep it free for its regular activities. The data will be fetched from a collection called "devices" in the batches of 2000 documents . This should further reduce read operations overhead.

Relevant Code

client = MongoClient('mongodb://H3-hostname:27017/', replicaset='rs0', readPreference='secondary')
.
.
.
.
.
devices = devices_collection.find({"status" : "ACTIVE"},{"key" : 1, "username" : 1}, sort = [("key", pymongo.ASCENDING)]).batch_size(2000)

Also tried creating the client in a little different way by mentioning all the hosts in the cluster as below

client = MongoClient('mongodb://H1-Hostname:27017, H2-Hostname:27017, H3-hostname:27017/?replicaset=rs0', readPreference='secondary')

Problem

As per my understanding, the above client will read ONLY from the secondary replica set. But when I run the script and look at the cloud manager, H1 which is primary shows page faults increased from ~15-20 to ~600 with a sharp jump.

Environment

• Mongo DB 3.0.3 (MMAPv1 storage engine)
• Python 2.7.9
• Pymongo 3.0.3
• OS Amazon Linux

Answer 1

I also posted this problem to mongo-user group, where it was answered.

First of all, the version of pymongo was wrong. It was 2.8, at least the one which python27 was using

$ python27
>>> import pymongo
>>> pymongo.version
    '2.8'

and

$ python
>>> import pymongo
>>> pymongo.version
    '3.0.3'

The appropriate syntax for that was

MongoClient('mongodb://H3-Hostname:27017/?readPreference=secondary')